Moon-eyes-fires-oceans: writing numbers as poetry

# Moon-eyes-fires-oceans: writing numbers as poetry

How do you publish a table of astronomical constants in a civilization whose scientific literature is composed entirely in metrical verse? Digits don't scan. Sanskrit's answer was to make the numbers out of *images*:

**moon** = 1 (there is one moon) · **eyes** = 2 · **fires** = 3 (the three ritual fires) · **oceans** = 4 · **arrows** = 5 (the love-god's five) · **seasons** = 6 · **sages** = 7 (the Seven Rishis) · **sky** = 0 (empty space).

String the words in place-value order and any number becomes a phrase. The system — *bhūta-saṁkhyā*, "object numerals" — was the working notation of Sanskrit science for over a millennium. Datta and Singh's standard history states the reason with textbook directness:

> "The ancient Hindu mathematicians and astronomers wrote their > works in verse. Consequently they strongly felt the need for a > convenient method of expressing the large numbers that occur so > often in the astronomical works… **The word numerals were invented > to fulfil this need and soon became very popular.**" > > — Datta & Singh, *History of Hindu Mathematics*, Vol. I (1938)

How it works

Two rules make the system precise rather than picturesque.

First, **place value**: each word is one digit, and the words run in a fixed order — by the convention *aṅkānāṁ vāmato gatiḥ*, "the digits run leftward," the *first* word is the units place. So *candra-netra-agni* (moon-eyes-fires) is not 123 but **321**.

Second, **synonym freedom**: any word meaning the object works. Datta and Singh's lists run to dozens of entries per digit — 1 is moon (candra, soma, śaśāṅka, indu…) or earth (bhū, dharā, kṣiti…); 2 is eyes, arms, wings, twins; 4 is oceans, Vedas, world-ages; 7 is sages, mountains, notes of the scale. Their worked example shows the number 1,230 written three different ways (*kha-guṇa-kara-ādi*, *kha-loka-karṇa-candra*, *ākāśa-kāla-netra-dharā* — the cached scan's "1,2.30" repaired to 1,230). As they put it, "the same number can be expressed in hundreds of ways… This property makes the word numerals specially suitable for inclusion in metre." A poet needing two long syllables reaches for one synonym; needing three shorts, another. The notation bends to the meter; the value never changes.

The result reads like surrealism and computes like arithmetic. A siddhānta verse stating a planetary parameter is simultaneously a grammatical Sanskrit sentence full of moons and oceans *and* a row in a data table — and that is precisely why India's astronomical constants could ride inside oral literature for centuries without corruption: [meter is a checksum](/c/9078a2cf-c645-5145-a8b8-41ec1fd1c89a), and a scribe who garbles a digit-word usually breaks the scansion or the sense.

The honest comparison

Symbol-laden number words are not uniquely Indian — Greek alphabetic numerals mapped letters to values, and gematria traditions played meaning-games with the results. But those are ciphers over an *additive* system, useful for dates and epigrams. Bhūta-saṁkhyā is different in the two ways that matter: it is **place-value** — the same short word-list composes numbers of any magnitude, exactly like digits — and it is **engineered for meter**, with synonym-richness as a design feature rather than an accident. It presupposes the positional idea whose Indian development [this corpus documents separately](/c/30ed9b53-ddb7-5087-820e-341220291d7b), and it coexisted with two sibling notations: Aryabhata's dense [alphabetic code](/c/b24b8d92-0600-5542-a179-8014ef059d75) (maximal compression, minimal poetry) and the later kaṭapayādi cipher (consonants as digits, whole meaningful words as numbers). Three solutions, one problem: making quantitative science publishable in verse.

The limitation is real and worth stating: word numerals are a *notation for stating* numbers, not for calculating with them. Computation happened on the dust board in digit numerals; the verse form was for transmission and memory. The division of labor is itself the insight — storage format and working format, separated a millennium before computing made that distinction routine.

Legacy

Bhūta-saṁkhyā stayed in continuous use as long as Sanskrit science did — Datta and Singh note it is "used even up to the present day" when numbers appear in Sanskrit verse, and inscriptions, chronograms and colophons across South and Southeast Asia carry dates in object-numerals. For historians, it is also a dating tool: a chronogram's imagery fixes a year as securely as a printed digit.

And it left a habit of mind. A tradition that spent a thousand years saying "moon-eyes-fires" for 321 was a tradition entirely comfortable with the idea that a number's *representation* is arbitrary and its *position* is everything — the exact conceptual ground on which [the world's decimal notation](/c/30ed9b53-ddb7-5087-820e-341220291d7b) grew. The poetry was never at war with the precision. The poetry was the file format.

---

Sources

- [Datta & Singh, *History of Hindu Mathematics: A Source Book*, 1938](https://archive.org/details/wg143) — Vol. I, pp. 54–56: the rationale, the 1,230 example, the synonym lists, and the "hundreds of ways" remark. - Plofker, *Mathematics in India*, 2009 — word-numerals and verse transmission (secondary synthesis, for context). - Salomon, *Indian Epigraphy*, 1998 — chronograms and dating usage (referenced for context only).

Related claims

- [Aryabhata's alphabetic number code (499 CE)](/c/b24b8d92-0600-5542-a179-8014ef059d75) - [The Yajurveda's ladder of tens (~1200–800 BCE)](/c/30ed9b53-ddb7-5087-820e-341220291d7b) - [The Līlāvatī: an algebra textbook written as poetry (1150 CE)](/c/9078a2cf-c645-5145-a8b8-41ec1fd1c89a)